Background

This was a project I developed for the Queensland University of Technology's Vice-Chancellor's Research Fellowship.

Description

The primary research question for this project was “What type of tools and techniques will allow users to explore, mine and analyse big spatiotemporal datasets?” The intent behind the exploration, mining, analysis would be to discover trends and patterns that support the work of users from non-computational fields, e.g. health and public policy. This is necessary because in this era of data-driven research, experts in specific domains, such as health and public policy, cannot make the most of their data without technical assistance. Simple tasks such as searching, interrogating, processing and publishing a dataset often require specific computer programming skills. The problem is made worse when the data is classified as being ‘big data’ and thus inherently too complex for any type of manual examination. This project aimed to address the aforementioned data engagement need.

In order to do so, the project aimed to research and develop new and innovative machine learning algorithms and couple them with data visualisation to construct a visual analytics system. The use case to which the visual analytics system would have been applied would have been to spatiotemporally analyse big socio-demographics dataset, such as the Australian census data. The analysis would have been conducted in the context of discovering the impact of societal prosperity on the health of the populace.

The use case was to involve data from the Australian Bureau of Statistics (ABS) and Cancer Council Queensland as the initial case study was to focus on cancer. However, secondary case studies in diabetes, neurodegenerative diseases, mental health and epidemics would have been conducted to demonstrate the applicability of the system to health data in general.

Outcome

Unfortunately, I was not successful with the post-doctoral fellowship grant for this project.